October 27, 2003

Greg Mankiw and Seatbelt Sam

"Huh," says the Thirteen-Year-Old. "Do you think this is true?"

He is on page 7 of Greg Mankiw's Principles of Economics. He reads:

Now consider how a seat belt law alters the cost-benefit calculation of a rational driver. Seat belts make accidents less costly for a driver because they reduce the probability of injury or death. Thus a seat belt law reduces the benefits to slow and careful driving. People respond to seat belts as they would to an improvement in road conditions--by faster and less careful driving.... Drivers who wear their seat belts are more likely to survive any given accident, but they are more likely to find themselves in an accident. The net effect is ambiguous.... [E]conomist Sam Peltzman showed that... [seat belt] laws produce... fewer deaths per accident and more accidents... little change in the number of driver deaths...

"Oh God!" I think to myself. First of all, Seatbelt Sam's regressions always seemed to me to be sensitive--the kinds of results you get when you are trying too hard and have argued yourself into believing the right specification is the one that is friendliest to your prior beliefs. Second, whoever is buckling or not buckling the seatbelt is not the rational utility maximizer of economics: the benefit-cost calculation for fastening your seatbelt is enormously large and positive for everyone, yet an appalling number of people did not fasten them back when Seatbelt Sam did his study. How are people who can't do the crudest benefit-cost calculations effectively supposed to do the math which leads them to say, "Aha! Because I am wearing my seatbelt I can drive three miles an hour faster on this curve"?

I can't think of an example more likely to turn off any skeptic--a more debatable illustration of the claim that "people respond to incentives." And that is too bad, because people do respond to incentives.

"Well," I say to the Thirteen-Year-Old. "Just suppose that Greg Mankiw is right here--that the introduction of seatbelts did lead to a worse world because the same number of drivers get killed in accidents and more pedestrians get killed." Then what should the government have done? If making cars safer is not good, then what would be good?"

"How about a large sharp spike right in the middle of the steering wheel pointing at the driver's stomach?" asks the Thirteen-Year-Old. "That would really encourage safe and careful driving!"

"Very good," I say. "I think that if you want to buy Seatbelt Sam's argument and call mandatory seatbelt laws examples of harmful and counterproductive government red tape, you'd better stand up and advocate their reverse--and the sharp spike in the middle of the steering wheel pointed at the driver's stomach seems a natural step. And I do note that neither Greg Mankiw nor Seatbelt Sam Peltzman has the courage of their convictions--so I don't think you need to take this example that seriously. That people will do what you give them an incentive to do is an important general point, however..."

Posted by DeLong at October 27, 2003 12:32 AM | TrackBack

Comments

Did the Thirteen-year-old come up with that "spike to the stomach idea" independently? That's amazing. I was reading Steven Landsburg's "Arm-Chair Economist" last year, and he said they should put on the stearing wheel a spike pointed at the heart -- almost the same thing!

Posted by: Bobby on October 26, 2003 08:59 PM

And of course, keeping the same safety rate while dramatically increasing speeds is nothing to sneer at.

Posted by: Jake McGuire on October 26, 2003 09:05 PM

On airline flights very young children normally fly sitting on parent's lap.

IIRC there was a movement a few years ago to legally require that such children fly in the equivalent of car seats, for safety reasons. This was OK with the airlines because it would require parents to buy an extra seat for each child.

The idea got scotched when somebody calculated how many parents would instead decide to travel with their families by car or bus rather than pay for the extra seat(s), factored in the higher risk of such driving compared to flying, and figured the new safety rule would probably cost lives on net.

Posted by: Jim Glass on October 26, 2003 09:09 PM

There is almost no end to the mind games you can play with this sort of thing. Some cities (Livermore) require sprinkler systems on all new residential construction. Does this encourage people to smoke in bed? If you look at the reduction in fire insurance costs, the net present value saving doesn’t pay for the sprinkler system. Do cholesterol lowering drugs discourage people from exercising and eating a healthy diet? I think the answer here is yes, but it seems very hard to prove. In any case we should require a license to practice regression analysis.

Posted by: A. Zarkov on October 26, 2003 09:23 PM

One summer long ago near a very bad local library, I read Sammy Davis Jr's biography, in which he said he lost his eye because the car manufacturer had put, for decorative purposes, a hollow decorative cone in the center of the steering wheel.

So, basically, it's been done.

Posted by: julia on October 26, 2003 10:23 PM

Of course the typical seat belt wearer "is not the rational utility maximizer of economics". But why is this relevant? A little instrumentalism is helpful here - people feel vaguely safer with seat belts on and on average drive in a more risky fashion. They don't have to do the cost/benefit calculus themselves, but the calculus can still accurately describe their behavior.

Also, it seems to me perfectly consistent for someone to both say that seat belts have an ambiguous effect on highway fatalities and that seat belts are a good thing (while a metal spike is not). Driving carefully is costly (slower, less fun, etc.), and seat belts allow us to reduce these costs (to a degree) with no greater danger of death.

Posted by: Ryan on October 26, 2003 10:47 PM

aw, c'mon, number crunching is fun. why, just the other day I downloaded California's Academic Performance Index data (http://www.cde.ca.gov/) just to make sure that my seat-of-the-pants beliefs about student performance versus parental education and income were reflected in the data. I was happy with the results; I doubt even an enterprising critic can make them go away.

back to the topic, I find the seatbelt conclusions difficult to believe. drivers today seem, on the whole, to drive much less unsafely than those with whom I remember riding in my youth. since I dislike a hypothesis that strikes me as willfully contradicting observation, it seems to me that if there are more accidents now, the simplest explanation is that traffic density is a lot higher, too.

I wonder when you start explaining to the thirteen-year-old that naive statistical tests of multivariate regression results are not to be trusted on their face.

Posted by: wcw on October 26, 2003 10:54 PM

Is the premise actually true? If I've had a glass of wine at supper, I drive more carefully, but that doesn't necessarily mean I'm less likely to crash. When I wear a seatbelt (nearly always, since many years) I feel more connected to the car, I have better posture - I'm a better driver. In my case there's a strong correlation between length of trip and likelihood of seatbelt use - and the short trips are likely the most likely to involve accidents (though minor ones). To disentangle these effects would require a lot of work.

On the other hand, the spike suggestion doesn't impress me - you'd have to show me what the fear/carelessness/unavoidable-accidents/
trip-length/speed/danger surface looks like to convince me - maybe the awareness of the proximity to a hard windshield is enough.

Posted by: rilkefan on October 26, 2003 11:04 PM

Good post, again! And I happen to agree fully to the opinions in it. We have this continously ongoing debate on traffic safety in median in the political and everyday debate etc. My impression from it is that the seatbelt is indeed a no-brainer, as is the ban on drunk driving. However, there seems to be slightly adverse effects from safety improvements, they tend to make people drive faster so that the potential from theimprovements, in terms of decreased total damage, is not reached to 100%.

Two additional points:

1) In risk assessments we display on important psychological bias (according to Akerlof?) - those involved in e.g. everyday driving underestimate its risk, just to feel better and to avoid cognitive dissonance. It should hence *not* be a good example of rationality.

2) There was a scientific result out in the media. Reserchers claimed that the percentage of violent deaths caused by accidents have been constant through millenias (N*1000yrs). Subconsciously, we should then be excellent, according to this, in judging risks.

Posted by: Mats on October 27, 2003 12:25 AM

".....drivers today seem, on the whole, to drive much less unsafely than those with whom I remember riding in my youth....."

If there's one thing I've learned it's that any observations that involve "back when I was a kid" - including my own - should be handled with hot tongs.

Posted by: Jason McCullough on October 27, 2003 12:57 AM

To cut accidents you need to enforce speed limits as well as seat belts. So what? That’s exactly what has been happening in Europe and deaths have fallen steadily. This big trend outweighs Seatbelt Sam’s one study.

People belt up to cut the risk from their own driving mistakes and from those of others – the latter factor justifies our intuition that the spike idea is wrongheaded. The risk-neutral moral imbeciles of neo-liberal economics, zero-weighting risks to others, may speed up to maintain constant risks to themselves. The prudent citizens of Adam Smith, driven by enlightened self-interest, will belt up to reduce risks to themselves and others since the belt has changed the risk production function. In a mixed society of moral imbeciles and citizens, overall risk falls. Thank God, a lot of people do keep rented cars clean, pick up others’ litter, give not sell blood, leave tips in restaurants they will never revisit, and so on.

Posted by: James on October 27, 2003 02:13 AM

I've always understood the notion of rational utility maximization to mean that groups of people behave as if they were rational utility maxiimizers, not that individual people consciously analyze the implications of behavior. For example, behavioral economics shows that individuals are not very rational, yet markets behave as if they were.

I've read enough reports of similar studies of auto and other safey to believe the reported result (directionaly - the exact magnitude is likely unknowable). For example, IIRC, a number of years ago there were German and Swedish studies of antilock braking that found improved safety in an accident was offset by less safe driving leading to more accidents.

Posted by: richard on October 27, 2003 04:30 AM

Jason,

I agree, in general, with you observation about "back when I was a kid" data. However, during a recent trip to the old homestead, my mother got onto this very topic. She remembers the driving in the 1940s and 1950s pretty well, and claimed there has been a simply massive reduction in accidents per driver. I'm not in the habit of questioning her memory about such things (don't want to reenforce her complaints about a failing memory), but she happened to have been going through old clippings and had at least one bit of evidence. My father's home town, at the time the article was written at town of 5000 inhabitants (not households - inhabitants) had already recorded 30 auto accidents of serious enough nature to end up in police records. That was in June. My first auto accident came when I was 1 year old. Apparently, nobody in my home town thought that was odd at the time.

Posted by: K Harris on October 27, 2003 04:33 AM

""How about a large sharp spike right in the middle of the steering wheel pointing at the driver's stomach?" asks the Thirteen-Year-Old. "

D*mn! I thought of that a while back. I was going to call it a 'Peltzman' (after the writer of such an article).


Posted by: Barry on October 27, 2003 04:37 AM

The thing that startles me is that such a dubious example is on page 7 of an introductory Economics book.

If economics students are actually being required to look at an example this early on in their introductory course and think "it doesn't matter if this example works in the real world, what matters is that I get the method of incentive-based thinking right" then we are in trouble. As the posts above illustrate, you can make an argument for anything if you take this route.

Posted by: Tom Slee on October 27, 2003 05:28 AM

To follow up Mats' point:

"However, there seems to be slightly adverse effects from safety improvements, they tend to make people drive faster so that the potential from theimprovements, in terms of decreased total damage, is not reached to 100%."

This is the way things work in all kinds of areas, not just incentives and economics. In physics, perturbation theory is the standard way to assess the impact of an external change to a system. The first order term assumes that the underlying system (in this case, the behaviour) does not change. The second order term describes the knock-on change from the system's response to the external shock. And this second order term is always opposite in sign to the first -- the perturbations dampen out. Same in chemistry (look up Le Chatelier's principle). Nevertheless, the direction of change (which is what we are interested in, of course, is given by the first order term -- that is, as if people did not change their driving behaviour.

There are changes that cannot be treated by perturbation theory of course, in which the series does not converge.

Posted by: Tom Slee on October 27, 2003 05:40 AM

Mankiw's comments do show the very narrow conception of psychology of some economists. It's pretty easy to put on a seatbelt, the kind of people who don't bother probably include more of the impulsive and risk taking types who drive less carefully. Also, there's a big incentive to avoid accidents because of the property damage. I agree with Brad, I'd be skeptical of any study that showed seat belts are counterproductive unless the evidence was overwhelming.

Posted by: Phil P on October 27, 2003 06:10 AM

>>Did the Thirteen-year-old come up with that "spike to the stomach idea" independently? That's amazing. I was reading Steven Landsburg's "Arm-Chair Economist" last year, and he said they should put on the stearing wheel a spike pointed at the heart -- almost the same thing!The Armchair Economist, so I cannot see how he would have run across it...

Posted by: Brad DeLong on October 27, 2003 06:17 AM

I suspect the question to ask your son is how
risky his behaviour would be while riding his
bicycle or roller blades if he didn't have a
kelmet on. Or, does he know anyone he thinks
might act differently without a helmet . I
remember reading a news story that suggested that
severe head traumas had not been reduced due to
helmet protection. I drew the implication that
kids were taking more risks and getting into ever more vioent accidents.

Posted by: malcolm on October 27, 2003 06:18 AM

There may be a bit of spurious correlation in the auto and bicycle data (see Malcolm's comment). The fad of bicycling in the style remeniscent of the Chinese circus may not be the direct outgrowth of greater helmet use, but I'd bet my last Crispy Creme it has contributed to injuries. Similarly, the return of the muscle car, and the associated driving practices, seems as much driven by fad-following as seatbelts. I remember being simply wowed as a kid when a guy at a gas station popped his hood and there, in all its glory, was a 426 hemi monster, all chrome and glory. Ads for hemis are back on the tube, and so, I understand, is the fad of street racing. "Road rage" is just an excuse for being a moron, but there seems to be a heck of a lot more high speed tail gating now than I remember from years back (there's that "when I was a kid" problem again).

Posted by: K Harris on October 27, 2003 06:35 AM

This the story I use in class to illustrate that "incentives matter": http://www.snopes.com/sports/soccer/barbados.htm. (And for more advanced classes, it illustrates the Lucas Critique.)

Posted by: Andrew John on October 27, 2003 07:03 AM

Sorry, that URL should be
http://www.snopes.com/sports/soccer/barbados.htm

Posted by: Andrew John on October 27, 2003 07:12 AM

With a seat belt law and seat belts in almost all cars, the ones who refuse to wear them are probably also the most reckless drivers. That's my hunch and nothing more than that. If true, how does it affect the statistics?

As I understand, traffic fatalities per mile have been going down pretty steadily, and both speed limit laws and seat belt laws have been shown to have a positive effect. In other words, wasn't Mankiw's hypothetical mental experiment matched by a reality which made the example a sort of worthless alternative-universe case?

And Mankiw is one of the sharp guys in the profession? This is the kind of thing that leads me to say that "Econ 101" can be a negative force in society. I remember 20 years ago having an argument with somebody about using antibiotics in livestock feed and breeding superbugs. It turned out that his econ teacher had used this as an example of tradeoffs, with the implicit conclusion that increased meat production was probably worth the risk. It was really something the econ teacher knew rather little about, but he managed to pump a cliche anti-regulation meme out into the world of public opinion. (My friend never did go on to Econ 102. I wonder how many other free-marketer talking points he was infected with).

Posted by: Zizka on October 27, 2003 07:54 AM

"How are people who can't do the crudest benefit-cost calculations effectively supposed to do the math which leads them to say, 'Aha! Because I am wearing my seatbelt I can drive three miles an hour faster on this curve'?"

This seems unnecessarily snarky to me. We don't minimize risk, we optimize it, and each of us does so every day. That's why we tend to drive faster on a bright sunny day than on a dark rainy night. And we do this without econometric models.

I don't know what Peltman concluded about the efficacy of seatbelt laws, though I'd expect seatbelt laws to make a positive difference. But people don't behave like crash test dummies, and it's perfectly sensible to take that into account. It's a good idea to examine the efficacy of particular regulations, even (particularly) when that conflicts with established orthodoxies.

Posted by: Jim Clark on October 27, 2003 07:55 AM

I have not seen the data that Sam uses, but isn't he reversing cause and effect? Increased accidents should stimulate increased seat belt use (among rational drivers) and not the other way around. Correlations do not determine cause and effect. As the denisty of vehicles increases the probability of accidents increase. Seat belt use and use laws are a response are an effect of the increase in accidents, not the cause of more accidents.

"Thus a seat belt law reduces the benefits to slow and careful driving. People respond to seat belts as they would to an improvement in road conditions..."

This argument is flawed, because it only includes the health risk to the driver. It does not cover vehicle loss and all the negatives associated with that (insurance never pays the whole cost, it is a pain to be without your car while it is being repaired, all the paperwork is a hassle, etc. etc.). Seat belts give added protection to the driver and use should go up when accident rates go up. However, seatbelts do NOTHING to change damage to the car. To assume that drivers care only about personal safety and nothing about the vehicle is gross oversight. Very few teenagers dread being killed in an auto accident because they don't think it will happen to them. Most teenagers live in mortal dread of having to call home and say, "Dad I wrecked your car."

Increases in traffic density can cause increases in accident rates in the absence of changes in driving habits. "Slow and careful driving" does not necessarily optimize your risk. Decreasing the speed differential on highways and finding pockets of lower car density do.

http://www.ibiblio.org/rdu/sae-sl85.html
http://www.ibiblio.org/rdu/speedsci.html

The example and the analysis is flawed. One wonders why Mankiw could not find a better example for his book.

Posted by: bakho on October 27, 2003 08:12 AM

"How are people who can't do the crudest benefit-cost calculations effectively supposed to do the math which leads them to say, 'Aha! Because I am wearing my seatbelt I can drive three miles an hour faster on this curve'?"

This seems unnecessarily snarky to me. We don't minimize risk, we optimize it, and each of us does so every day. That's why we tend to drive faster on a bright sunny day than on a dark rainy night. And we do this without econometric models.

I don't know what Peltman concluded about the efficacy of seatbelt laws, though I'd expect seatbelt laws to make a positive difference. But people don't behave like crash test dummies, and it's perfectly sensible to take that into account. It's a good idea to examine the efficacy of particular regulations, even (particularly) when that conflicts with established orthodoxies.

Posted by: Jim Clark on October 27, 2003 08:34 AM

in re: Peltzman, I didn't find a good data source with which to play, but I did find the following paper based on Canadian data (which the author prefers):
http://economics.uwaterloo.ca/sen/traffic%20paper-RESTAT.pdf

Posted by: wcw on October 27, 2003 08:43 AM

When I was a graduate student I knew a lot of economics graduate students --- these were people who had got places on the pretty competitive M Phil in economics at Oxford, i.e. some way beyond Econ 101 --- and nearly every one of them would solemnly put forward the spike-pointed-at-the-heart solution to road safety when the topic came up. Without any empirical doubts (or indeed concerns about what would happen to social welfare if everyone was too scared to drive), too. It's taken a long time for me to learn to take economists seriously...

Posted by: Nasi Lemak on October 27, 2003 09:06 AM

I think Peltzman's mistake was to jump from "seat belts make people drive more recklessly" to "seat belts aren't, on balance, a good thing." The second point doesn't follow from the first, since there's no doubt that seat belts reduce the risk of injury and death to drivers.

But this also means that citing reduced traffic fatalities per mile doesn't speak to the question of whether seat belts make people drive more recklessly. And there is considerable evidence that Peltzman's broader point -- that safety improvements make people drive more recklessly -- was right. In a wide range of countries where seat-belt laws were introduced, the rate of traffic injuries suffered by pedestrians and cyclists rose significantly.. There's an interesting study suggesting that improvements in highway design -- wider roads, better surfaces -- have actually increased accident rates. And then there are the empirical studies of people's behavior. The most famous is the one of the German cab drivers. Half of the drivers in a fleet of taxis in Munich were given cars equipped with anti-lock brakes. The other half weren't. Their driving behavior was studied -- unbeknownst to them -- for three years. The drivers in the cars with ABS became significantly more aggressive and worse drivers. And even though the ABS should have reduced their accident rates, it didn't.

Again, though, the conclusion that seatbelt laws are a bad idea doesn't follow. At this point, I don't think anyone can argue that seatbelts reduce total fatalities -- even if you take into account all the other things that affect fatality rates (including improvements in medical technology, changing demographics, drunk-driving rates, etc.).

I also wonder if the risk-compensation effect doesn't wear off over time. After all, if you're a young driver, you've always driven with seat belts (at least they've always been required and in the car). So would seat belts really make you more aggressive? I disagree with Brad about the need for people to make elaborate cost-benefit calculations for Peltzman's theory to hold. But I also find it implausible that someone who's always been seatbelted would say to himself "Hmmm. . . If I didn't have to wear a seatbelt, it would be more dangerous for me and I would drive more slowly. So, since I do have to wear a seatbelt, I will drive faster."

In that respect, it's interesting that, in Canada at least, pedestrian and cyclist injuries declined over time after rising in the years just after the seat-belt law was passed.

Posted by: James Surowiecki on October 27, 2003 09:11 AM

This does seem to sound like an inept example for an Econ 101 text book *for this point*, but I do think it is useful to point out that policy decisions can have unforeseen consequences on other people's decision-making. So, we might all remember that an important part of the debate surounding the proposed increase in the speed limit on interstates from 55 to 70 (or greater) was that increasing the speed limit could lead to a large increase in traffic fatalities. This large increase did not materialize, and there may have even been a small decrease.

I think this qualifies as a legitimite surprise, although in retrospect, we can identify factors that could have offset the easily understood dangers of having a very high speed limit. One factor is that higher speed limits might have decreased the variability in speeds on the highway, since variability is what kills you. Another factor is that increased speeds (and therefore potentially increased capacity) on the freeways made the rural and country roads less attractive routes; this turns out to be safer insofar as the oncoming traffic on a 2-lane country raod poses a much larger threat to safety than a 10 pmh increment in speed on a 4+ lane divided highway. (This second point was completely lost to me until I moved to central Missouri, where it seems to be absolutely true.)

Now the real fun for me is that I bet if you asked drivers what *their* goals were in switching from country road to superhighway, most of them would give just the argument about decreased travel time, or travel time plus a reduced chance of getting caught in a speed trap. The effect on safety then almost seems like one of those invisible hand things.

Posted by: Jonathan King on October 27, 2003 09:28 AM

the spike idea will hurt you if some idiot rams you from the front.

Posted by: eric bloodaxe on October 27, 2003 10:30 AM

To me the significance is not that a not-very-good example was chosen, but that the example suggested or required jumping to probably-wrong conclusions about a real public issue. The same would be true of my antibiotic-in-feed example. There's a kind of unreal Utopianism in economics which (combined with the academic tendencies toward formalization, hypotheticals, and "possible worlds" thinking, has led me to Nasi's conclusion. Economics should not be taken on an empty stomach or without adult supervision.

Posted by: Zizka on October 27, 2003 11:03 AM

I think James Surowiecki's point is quite salient, and wonder whether later in the book Mankiw uses this example to discuss the analysis of such claims (leading to a discussion the elasticity of demand for safety, external effects, etc). My intuition tells me that seat belts will cause people to drive less safely, but not enough to completely counteract the benefits of being belted while in an given accident. The anti-seat belt law argument still hinges on the notion that that isn't a decision that the government should be making for people.

FWIW, on the "having grown up with seat belts will prevent drivers from being able to make such comparisons" point: introspection tells me that I would attempt to drive significantly more carefully without a seat belt because, to use the words of another commenter, I don't feel as connected with the car without a seat belt. Extensive experience driving without wearing seat belts would probably reduce that feeling. --sw

Posted by: Scott Wood on October 27, 2003 11:08 AM

I think James Surowiecki's point is quite salient, and wonder whether later in the book Mankiw uses this example to discuss the analysis of such claims (leading to a discussion the elasticity of demand for safety, external effects, etc). My intuition tells me that seat belts will cause people to drive less safely, but not enough to completely counteract the benefits of being belted while in an given accident. The anti-seat belt law argument still hinges on the notion that that isn't a decision that the government should be making for people.

FWIW, on the "having grown up with seat belts will prevent drivers from being able to make such comparisons" point: introspection tells me that I would attempt to drive significantly more carefully without a seat belt because, to use the words of another commenter, I don't feel as connected with the car without a seat belt. Extensive experience driving without wearing seat belts would probably reduce that feeling. --sw

Posted by: Scott Wood on October 27, 2003 11:12 AM

I agree with Nasi and Zizka about
"Economics should not be taken on an empty stomach or without adult supervision." The scary thing here is that a textbook is meant to *be* adult supervision.

However, as for Nasi's point about Oxford M.Phil's, I suspect their solemn advocation of the spike comes more from an Oxbridge tendency to want to debate everything and to play the devil's advocate no matter what position others take.

Not that that makes me feel more comfortable to have our future in these people's hands....

Posted by: Tom Slee on October 27, 2003 11:14 AM

Maybe I won't encourage my 12-year old to read econ...

I think there's a reasonable rational choice story in which seat belt laws lead to safer driving. The gist is that different groups, Citizens and Rebels, try to signal their status. A law allows the Rebels to signal more cheaply by not buckling up rather than driving badly, and motivates the Citizens to do more to show their prudence than just buckling up. The full story is at my blog, http://eastmania.blogspot.com

Posted by: Wayne Eastman on October 27, 2003 11:43 AM

Maybe I won't encourage my 12-year old to read econ...

I think there's a reasonable rational choice story in which seat belt laws lead to safer driving. The gist is that different groups, Citizens and Rebels, try to signal their status. A law allows the Rebels to signal more cheaply by not buckling up rather than driving badly, and motivates the Citizens to do more to show their prudence than just buckling up. The full story is at my blog, http://eastmania.blogspot.com

Posted by: Wayne Eastman on October 27, 2003 11:45 AM

Easy on the econ-bashing (NB Nasi and Zizka). Sometimes economic logic leads to outlandish or blatantly wrong statements, but the antidote to a bad model or bad statistics is a good model, not dogmatism. It's usually a better bet that if someone doesn't follow your model, your model's wrong, not that he's stupid. Or your statistics are bad.

What we have here is an empirical question--how much do seatbelts affect the safety of driving? If people are in the least bit rational AND if they're in the least bit selfish, they'll drive more dangerously. So this part is an argument about rationality and self-interestedness in theory. In terms of statistics, there's the endogeneity problem again--seat-belt wearing is a function of safety, and safety is a function of seat-belt wearing. There are ways to deal with this (instrumental variables--say, using seat-belt laws or fines as instruments) but caution has to be used, and it's easy to fall into a trap of data mining.

(My own guess is that driving deteriorates quite a bit and fatalities fall just a little. But I need good data to back this up and I have to be prepared to change my mind if I think that I'm wrong.)

It's interesting--normally the left is quick to point out moral hazard in markets unless the right points it out first (for once). In that case, "better to fight than be right?" Now what about moral hazard in building in fire-prone areas, if the government's always on hand to bail people out?

Posted by: Chris on October 27, 2003 12:54 PM

Tom: No. They didn't want to argue for the sake of a an argument that the spike/steering wheel venture would be successful - rather they saw it as a great insight of their subject and a piece of really original thinking.

Chris: if presented with vast amounts of actual data, some experimental evidence, and some serious historical insight, then I might accept that there was a serious safety argument against seatbelts. However the spike/steering wheel thing is a good example of what Green and Shapiro called "stylised facts": you start off with the theory, you illustrate the theory with an account of reality which is poor or untested, and then you use the account of reality as evidence to back up your theory. In this case it becomes a "stylised fact" that seatbelts are bad for road safety. I think this kind of thinking is very bad for social science inquiry and unfortunately it characterises quite a lot of economics teaching ime.

Posted by: Nasi Lemak on October 27, 2003 01:25 PM

Nasi -- now you are scaring me.

Posted by: Tom Slee on October 27, 2003 01:47 PM

My argument wasn't with the actual content of economics. It was with two cases I knew of now in which introductory economics courses rather gratuitously used controversial, probably-fallacious hypothetical examples to make their (anti-regulation) points. To me there's something bizarre and offensive about that, but it is the kind of hyper-rational, counter-intuitive, contrarian thing that economists delight in. My conclusion is that when economists are right, they're very right, and that when they'r wrong, they're very wrong, and that they themselves aren't usually able to tell which is which, or even care very much. Sort of out-of-control snarkiness.

I feel pretty much the same way about analytic philosophers, quantum-physics metaphysicians, and post-modernists. Counterintuitive snark ingeniously and powerfully argued for its own sake. But economists are much more influential than the others.

There's a storefront church here in Oregon which makes opposition to seatbelt laws central to its political theology (along with the rest of the religious right stuff). Somehow it seems more obnoxious when a church does it.

Incidentally, there are facts on this case, I'm pretty sure. I guess we're in hypothetical-possible-universe-land, since no one seems to have mentioned them so far.

Posted by: Zizka on October 27, 2003 01:57 PM

I just Googled around and found Green and Shapiro's "Pathologies of Rational Choice Theory". Thanks Nasi!

Posted by: Zizka on October 27, 2003 02:26 PM

"Did the Thirteen-year-old come up with that "spike to the stomach idea" independently?"

Brad DeLong responds, "That's amazing. I was reading Steven Landsburg's "Arm-Chair Economist" last year, and he said they should put on the stearing wheel a spike pointed at the heart -- almost the same thing! The Armchair Economist, so I cannot see how he would have run across it..."

Circa 1980, I attended a guest lecture by Gordon Tulloch at Va Tech (D-oh! There goes the National Championship!), in which Dr. Tulloch noted that the senseless slaughter of pedestrians could be greatly reduced by placing a spike at the center of steering wheels. (I remembered it, and have since related it many times, because I thought it was a really cool example of economics yielding clever advice. ;-))

So the ol' spike-in-a-steering-wheel idea is (at least) approximately a quarter century old.

But your 13 year-old shouldn't be discouraged. While at Va Tech, I invented the "space elevator" idea, after having read Robert Heinlein's "The Moon is a Harsh Mistress."

Only later did I discover that Arthur C. Clarke had stolen my idea several a couple years earlier.

P.S. Hmmmmm...that's eerie! According to this website, Arthur C. Clarke first proposed the space elevator idea in a 1979 book. That might actually have been a year or two *after* I read "The Moon Is a Harsh Mistress." So maybe Arthur C. Clarke stole the space elevator idea from me! ;-) (That b@$$+^%!! ;-))

http://www.hindustantimes.com/news/181_374838,00040005.htm

P.P.S. D-oh! Apparently, I unknowingly stole the idea from Jerome Pearson, who wrote about it in 1975. (I don't think I read "The Moon is a Harsh Mistress" until after 1976.) But Jerome Person apparently stole it from John Isaacs in 1966, who stole it from Yuri Artsutanov in 1960, who stole it from Konstantin Tsiolkovsky...

http://science.nasa.gov/headlines/y2000/ast07sep_1.htm

...in 1895! (Whoa! Can't touch that! :-))

Posted by: Mark Bahner on October 27, 2003 02:29 PM

"the spike idea will hurt you if some idiot rams you from the front."

No problem. Automotive engineers will come up with a sensor and supporting software to decide whether one is about to be rammed from the front by some idiot, or whether one is about to knock off a defenseless pedestrian. (Who might also be an idiot. That would complicate the design. ;-))

For the former, the airbag deploys, for the latter, the spike. (Not sure which would deploy if subject pedestrian was an idiot.)

Posted by: Mark Bahner on October 27, 2003 03:03 PM

"Circa 1980, I attended a guest lecture by Gordon Tulloch at Va Tech (D-oh! There goes the National Championship!), in which Dr. Tulloch noted that the senseless slaughter of pedestrians could be greatly reduced by placing a spike at the center of steering wheels."

Steering columns used to be extremely dangerous, spike or no. Legal requirements were put in place that required steering columns to be less dangerous--these days, they are collapsible. Anyone know any studies on the impact of these on driver behavior?

More generally, I doubt that the presence of safety technology necessarily leads to carelessness.

"To me there's something bizarre and offensive about that, but it is the kind of hyper-rational, counter-intuitive, contrarian thing that economists delight in."

I get the feeling that a fair number of economists (I don't mean you, Prof. Delong) like the feeling they are superior to the common herd. It also meshes well with the beliefs of the nastier sort of right-wing economists, who often seem to believe in something very much like an aristocracy, to which they fancy they belong. (In a real aristocracy, most such people discover they do not belong.) Myself, I have found that such knowlege, when I in fact have it and am not being an egotistical jerk, is a heavy burden. I would prefer not watching people rationalize doing awful things to themselves.

Posted by: Randolph Fritz on October 27, 2003 03:12 PM

wcw wrote, ".....drivers today seem, on the whole, to drive much less unsafely than those with whom I remember riding in my youth....."

Jason McCullough responded, "If there's one thing I've learned it's that any observations that involve "back when I was a kid" - including my own - should be handled with hot tongs."

This reminds me of a apparently unrelated matter. One thing that fascinated me about the cold fusion brouhaha in 1989 was that theoretical physicists could come up with a theory, no matter whether cold fusion was real or fake.

So here is my theory for why wcw's observation is indeed *correct,* but is simultaneoulsly *incorrect,* based on my own life:

1) When I was a passenger (with people my age) in high school and college, the drivers were indeed lunatics.

2) Now that I don't ride in cars driven by high school and college kids, I don't ride with lunatics.

3) BUT...today, if I drive near local high schools or the University of North Carolina or Duke, I see that high school and college kids STILL drive like lunatics.

So...wcw is both correct and incorrect. QED. ;-)

Posted by: Mark Bahner on October 27, 2003 03:22 PM

"Steering columns used to be extremely dangerous, spike or no. Legal requirements were put in place that required steering columns to be less dangerous--these days, they are collapsible. Anyone know any studies on the impact of these on driver behavior?"

I doubt one could tease out the effects of the the many, many changes that have occurred since...say, 1965:

1) Seat belts, front and rear (I think the rear seat belts didn't happen until the mid-to-late 1960s),

2) Air bags...first driver, then passenger,

3) Stricter drunk driving laws,

4) Lighter cars, starting in the mid-1970s...but now somewhat reversing, so in fact the weight *difference* between vehicles is probably actually higher than in the 1970s,

5) Collapsible steering wheels,

6) Declining power from the 1960s to 1980s...but power coming back in the 1990s,

7) Replacement of station wagons (this will date me, but when I was growing up, kids piled into the backs of station wagons, where there were of course no seat belts) with minivans and SUVs,

8) Interstate speed limits going down, then coming back up,

9) Third lights in the rear (really does help), and daytime driving lights in the front (definitely helps on some days),

10) More cars on the road per mile than in 1965 (i.e., the total number of miles driven has risen faster than the number of miles of road built),

11) Average age of drivers increasing...

...and those 11 are just the start of a list. I forgot anti-lock brakes...front-wheel drive...

P.S. From personal experience, I've noticed that if I watch a Bond movie at a theater, I definitely drive home like a lunatic. But since Bond movies have been around since 1965, I guess we don't have to correct for that. ;-)

Posted by: Mark Bahner on October 27, 2003 03:47 PM

Another example of econs, lawyers, et. al., that think they know something can be seen in Brad and the blogosphere's continuing championing of outsourcing.

I am for Free Trade and I am against tariffs, but I think it is reasonable to look (a new) at some of the basic assumptions:

A) Is a job that moves from the US to a country that regularly violates labor and pollution controls a free trade,

B) Is it true that jobs skills are fungible? Does comparative advantage mean that it is okay for the US to get rid of jobs in steel and metal or would a realistic/game theory look at the world say we should do something different even if comparative advantage argues differently.

pejman is currently extolling how wonderful GM foods are, a gift to the world he says, but deletes comments discussing Monsanto, ready-roundup seeds, copyright issues and Monsanto suing farmers, or comments that mention that scientists still have no real understanding of what 80% of DNA does, and maybe that is a warning flag. Pejman is truly an idiot lawyer.

Posted by: bangalore on October 27, 2003 04:40 PM

Another example of econs, lawyers, et. al., that think they know something can be seen in Brad and the blogosphere's continuing championing of outsourcing.

I am for Free Trade and I am against tariffs, but I think it is reasonable to look (a new) at some of the basic assumptions:

A) Is a job that moves from the US to a country that regularly violates labor and pollution controls a free trade,

B) Is it true that jobs skills are fungible? Does comparative advantage mean that it is okay for the US to get rid of jobs in steel and metal or would a realistic/game theory look at the world say we should do something different even if comparative advantage argues differently.

pejman is currently extolling how wonderful GM foods are, a gift to the world he says, but deletes comments discussing Monsanto, ready-roundup seeds, copyright issues and Monsanto suing farmers, or comments that mention that scientists still have no real understanding of what 80% of DNA does, and maybe that is a warning flag. Pejman is truly an idiot lawyer.

Posted by: bangalore on October 27, 2003 04:43 PM

anyone here still reading to the end of the comments thread and nevertheless looking for seatbelt data, please backtrack to the paper I found. it has numbers, and even an argument (which I am probably not competent to critique) that the seatbelt coefficient is biased down due to simultaneity effects.

on the subject of ABS, however, my inexpert observation indeed yields confirmation. my wife is scared to fly but not to drive to work. seatbelts don't buttress the driver's illusion of control that results in that counterintuitive attitude; ABS, with its promise of skidless stops, will. to find ABS had a negative coefficient in a population of aggressive drivers (cabbies) would not surprise me in the least.

Posted by: wcw on October 27, 2003 05:02 PM

Mark--
The Space Elevator idea was actually investigated by NASA. The idea is far older than you think:

http://science.nasa.gov/headlines/y2000/ast07sep_1.htm

"As early as 1895, a Russian scientist named Konstantin Tsiolkovsky suggested a fanciful "Celestial Castle" in geosynchronous Earth orbit attached to a tower on the ground, not unlike Paris's Eiffel tower. Another Russian, a Leningrad engineer by the name of Yuri Artsutanov, wrote some of the first modern ideas about space elevators in 1960. Published as a non-technical story in Pravda, his story never caught the attention of the West. Science magazine ran a short article in 1966 by John Isaacs, an American oceanographer, about a pair of whisker-thin wires extending to a geostationary satellite. The article ran basically unnoticed. The concept finally came to the attention of the space flight engineering community through a technical paper written in 1975 by Jerome Pearson of the Air Force Research Laboratory. This paper was the inspiration for Clarke's novel."

Alison Sky & Michael Stone's book, *Unbuilt America* (1976-McGraw Hill) includes images from an old NASA study of the physics of such an elevator.

Posted by: James R MacLean on October 27, 2003 05:10 PM

Well, if you don't use that book to teach Economics, it might come in handy for teaching Critical Thinking.

Posted by: northernLights on October 27, 2003 06:54 PM

Have any of you actually read the Peltzman piece? When I taught HS economics a few years ago I used Mankiw's book and at the direction of my students I went to the original article(which Mankiw doesn't cite particularly well). I don't have a copy of the article with me, but if I remember correctly these were the conclusions:

After the enactment of seat belt laws:
1) Driver and passenger fatalities(per mile driven) went DOWN
2) Accidents(per mile) went UP
3) Pedestrian fatalities went UP
4) Total number of deaths(i think per mile driven) stayed relatively constant.

As an econ teacher I like this simple example, not because it is necessarily true, but because it is both plausible, and testable. I see economics as providing my students a framework within which to tell stories about the world and then design ways to test if their stories are true(or at least likely). This seems a whole lot better than simply telling stories based on personal experience with no logical underpinnings and without an understanding or willingness to question whether or not the story is actually true.

Posted by: ml on October 28, 2003 05:25 AM

ML: if a careful examination of the facts ends up showing that Pelzman was right, then he was right, and people should be told about it.


This is indeed a beautiful example of the workings of economics and the benefits of economic thinking -- if and only if it is true. Otherwise, it is just a made-up fact used to confirm the economists' prejudices and impose them on 101-level undergraduates who will, in most cases, never learn economics but probably will retain the example.

A "Mankiw + seatbelts" Google finds mostly intro-level courses, often with the phrase "How Seatbelts Kill" (which apparently was Mankiw's chapter title). Pelzman's article was from 1975, when seatbelt laws were under discussion and relatively new where they existed at all. (The role Pelzman's piece played in the argument can be guessed; seemingly, it was the kneejerk economist's role of explaining that government regulation of business for any purpose is always counterproductive and prohibitively expensive).

There's been a tremendous amount of data generated since then, and we should know a lot more than Pelzman did by now. According to this, Pelzman's results have been not been confirmed since 1975. It would ssem that, in conformance with normal scientific practice, the example should be dropped pending confirmation.

My opinion remains unchanged.

Posted by: Zizka on October 28, 2003 06:31 AM

Here are two references.

http://tinyurl.com/soqa

http://tinyurl.com/sooo

Posted by: Zizka on October 28, 2003 06:42 AM

Here are two references.

http://tinyurl.com/soqa

http://tinyurl.com/sooo

Posted by: Zizka on October 28, 2003 06:47 AM
Post a comment